22 comments

[ 3.2 ms ] story [ 59.5 ms ] thread
Whoops, the original implementation used microseconds it now uses seconds (with a max of ten).
Nice.

The "copyright" line is really tiny. Your twitter usernames are there, and I assume you want people to see them, so I'd make it larger.

I rarely notice copyright lines. The fact that you did makes me suspect that they succeeded.
Hello, I have no doubt that this tool is really usefull. But as a non techie I dont understand the purpose. Can somebody explain me or refer me to an article that can explain why test HTTP caches and what is it?
reallyslowsite.com is designed to be a tool that developers can use to simulate how their application responds to expensive HTTP requests.

You could use this, for example, to test that your mobile application is properly caching HTTP requests (it would be noticeably slower, even with a fast connection, because the request is guaranteed to take at least 2 seconds).

The specific use case I had in mind when I created it was testing the HTTP caching that will be used internally by the OpenStack "glance" project.

This site is slow :)

I don't know Ruby, but how does this work:

    get '/', :agent => /curl/ do
      (slow_thing, time) = slow_down(DEFAULT_SLEEP)
      "SLOWER THAN #{slow_ting}. Putting the brakes on for #{time} seconds.\n"
    end
It binds slow_thing and then references it as slow_ting (without the h) in the interpolation? What?

It appears twice so it must be intentional?

Oops, that's a typo in code that hasn't been deployed yet. Our regression suite missed it ;-)
Ah, makes sense, you just get the empty string instead of "turtles marching through peanut butter".

I am used to Perl where your code won't compile if you reference an undefined variable in a string interpolation.

This is one of the things that node.js is really good at, and it's interesting that the sample code on the front page of nodejs.org was pretty similar to this app. Here's the rough equivalent in node:

  require('http').createServer(function (request, response) {
    response.writeHead(200, {'Content-Type': 'text/plain'});
    var time = Math.max(0, Math.min(60000, parseInt(request.url.slice(1)) || 0));
    setTimeout(function() { 
      response.end('This page load took ' + time + 'ms\n'); 
    }, time);
  }).listen(8124);
(They have since changed it, and I suspect that it was because new users were confused as to why their supposedly "fast" server always took 2 seconds to respond to requests)

(edit: code formatting)

> This is one of the things that node.js is really good at

Why does every simple task motivate people to show some kind of "hello world" code in their favorite language and framework?

This web service is just a tiny exercise that is easily realized with a few lines in any language (even C or Bash) using any web server. In particular, I don't see in how far this code demonstrates the strengths of node.js.

Amusingly, my involvement came from showing shmeedog how nice his service would look in Sinatra versus WSGI.
Perhaps because, though Sinatra is actually one of my favorite frameworks, it doesn't work very well for this. The author comments that it's limited to 10 seconds because each connection is handled by a blocking Thin. Eventmachine could also solve this problem, in a similar amount of code.

And I was just pointing out that this happened to be the official hello world for node.js.

shmeedog wrote a server in eventlet that solves the blocking thin problem.